View
10
Download
0
Category
Preview:
Citation preview
1
Cationic Lipid-DNA Complexes for Gene Therapy:
Understanding the Relationship between Complex Structure
and Gene Delivery Pathways at the Molecular Level
Kai Ewert(1), Nelle L. Slack(1), Ayesha Ahmad(1), Heather M. Evans(1), Alison J. Lin(1),
Charles E. Samuel(2), and *Cyrus R. Safinya(1)
(1) Materials Department, Physics Department, and Biomolecular Science and
Engineering Program, University of California, Santa Barbara, CA 93106-5121
(2) Molecular, Cellular, and Developmental Biology Department, and Biomolecular
Science and Engineering Program, University of California, Santa Barbara, CA 93106-
9610
*Corresponding author:
Tel: (805) 893 8635, Fax: (805) 893 7221, e-mail: safinya@mrl.ucsb.edu
Key Words: gene therapy, gene delivery, nonviral, small-angle x-ray scattering,
multivalent, cationic liposomes, lipofection
2
Abbreviations:
BSA Bovine serum albumin
CL Cationic liposome
DMEM Dubellco’s modified Eagle’s medium.
DMRIE 2,3-Di(myristyloxy)propyl(2-hydroxyethyl)dimethylammonium bromide
DOPC 1,2-Dioleoyl-sn-glycerophosphatidylcholine
DOPE 1,2-Dioleoyl-sn-glycerophosphatidylethanolamine
DOSPA 2,3-Dioleyloxy-N-[2-(sperminecarboxamido)ethyl]-N,N-dimethyl-1-propylammonium chloride
DOTAP 2,3-Dioleyloxypropyltrimethylammonium chloride
HII
C Inverted hexagonal phase of lipid-DNA complexes
LSCM Laser scanning confocal microscopy
LαC Lamellar phase of lipid-DNA complexes
MVL5 N1-[2-((1S)-1-[(3-Aminopropyl)amino]-4-[di(3-aminopropyl)amino]butylcarboxamido)ethyl]-3,4-di[oleyloxy]-benzamide
NLS Nuclear localization sequence
PEG Poly(ethylene glycol)
RLU Relative light units
SAXS Small angle x-ray scattering
TE Transfection efficiency
XRD X-ray diffraction
κ Membrane bending rigidity
κGMembrane Gaussian modulus
σMMembrane charge density
3
ABSTRACT
Cationic liposomes (CLs) are used as gene vectors (carriers) in worldwide human
clinical trials of non-viral gene therapy. These lipid-gene complexes have the potential of
transferring large pieces of DNA of up to 1 million base-pairs into cells. As our
understanding of the mechanisms of action of CL-DNA complexes remains poor,
transfection efficiencies are still low compared to gene delivery with viral vectors. We
describe recent studies with a combination of techniques (synchrotron x-ray diffraction
for structure determination, laser-scanning confocal microscopy to probe the interactions
of CL-DNA particles with cells, and luciferase reporter-gene expression assays to
measure transfection efficiencies in mammalian cells), which collectively are beginning
to unravel the relationship between the distinctly structured CL-DNA complexes and their
transfection efficiency. The work described here is applicable to transfection
optimization in ex vivo cell transfection, where cells are removed and returned to patients
after transfection. CL-DNA complexes primarily form a multilayered sandwich structure
with DNA layered between the cationic lipids (labeled LαC). On rare occasions, an
inverted hexagonal structure with DNA encapsulated in lipid tubules (labeled HII
C) is
observed. A major recent insight is that for LαC complexes the membrane charge density
σM of the CL-vector, rather than the charge of the cationic lipid alone, is a key universal
parameter that governs the transfection efficiency of LαC complexes in cells. The
parameter σM is a measure of the average charge per unit area of the membrane, thus
taking into account the amount of neutral lipids. In contrast to LαC complexes, HII
C
complexes containing the lipid 1,2-dioleoyl-sn-glycerophosphatidylethanolamine (DOPE)
exhibit no dependence on σM. The current limiting factor to transfection by cationic lipid
4
vectors appears to be the tight association of a fraction of the delivered exogenous DNA
with cationic cellular molecules, which may prevent optimal transcriptional activity.
Future directions are outlined which make use of surface-functionalized CL-DNA
complexes suitable for transfection in vivo.
5
INTRODUCTION
The field of gene delivery by synthetic (non-viral) vectors continues to attract the
interest of a large number of research groups, motivated mainly by the promises of gene
therapy [1,2,3,4,5,6,7]. Following initial landmark studies [8,9,10], cationic liposomes
(CLs; closed bilayer membrane shells of lipid molecules shown in Fig. (1)) have been
established as one of the most prevalent synthetic vectors [2]. They are already used
widely for in vitro transfection of mammalian cells in research applications. To enable
the use of CL-DNA complexes in gene therapy, their mechanism of action is investigated
extensively in many laboratories, concurrently with ongoing, mostly empirical, clinical
trials, e.g. to develop cancer vaccines [1,2,3,6]. Most revealing, a current compilation
(Table 1) [11] of open gene therapy clinical trials worldwide shows that among 474
trials, over one quarter (126) are with non-viral methods. These are in turn nearly equally
split between lipofection-based vectors (including liposomes as well as mixtures of DNA
with lipid, polymer, and other small molecules described in the articles in this issue) and
naked DNA.
Insert Table 1.
A sense of urgency for developing efficient synthetic carriers stems from the recent
tragic events associated with the use of engineered adenovirus vectors. A patient died in a
clinical trial due to an unanticipated severe immune response [12]. This points to an
important advantage of CL-vectors over viral carriers, namely, the lack of a specific
immune response due to the absence of viral peptides and proteins. In addition, in
current trials using modified retrovirus vectors to treat children with severe combined
6
immunodeficiency (SCID), a French gene therapy team faced a major setback when one
patient (out of ten) developed a blood disorder similar to leukemia. The disease was
confirmed to have resulted from insertion of the transferred DNA into the initial coding
region of a gene related to the early development of blood cells [13]. Moreover, while
viral capsids have a maximum DNA-carrying capacity of about 40,000 base-pairs [5],
CL-DNA complexes place no limit on the size of the DNA, since the vector is formed by
self-assembly [14,15,16]. Thus, if complexed with human artificial chromosomes
[17,18], optimally designed CL-vectors offer the potential of delivering multiple human
genes and regulatory sequences extending over hundreds of thousands of DNA base-
pairs. In fact, fractions of an artificial human chromosome with a size of the order of 1
million base-pairs have been transferred into cells using cationic lipids as a vector, albeit
extremely inefficiently [17,18].
When solutions of DNA and cationic liposomes are combined, CL-DNA
complexes form spontaneously. This is schematically illustrated in Fig. (1). The entropy
gain associated with the release of tightly bound counterions from the DNA and the lipid
bilayers provides the main driving force for this process. The liposomes are made up of
at least two lipids, a cationic lipid and a neutral lipid (sometimes called helper lipid).
Depending on the ratio of charges on the cationic lipid and the DNA, anionic, neutral or
cationic complexes are obtained. We refer to neutral complexes, where the charges on
the DNA exactly match those on the cationic lipids, as isoelectric. For transfection,
positively charged complexes are used, as this promotes adhesive interactions with the
cell’s plasma membrane (see below).
Insert Fig. 1
7
In spite of all the promise CLs hold as gene vectors, their transfection efficiency
(TE), which is a measure of the amount of successfully transferred and expressed DNA,
remains low compared to that of viral vectors. This has spurred intense research activity
aimed at enhancing TE [1,2,3,4,5,6,7]. Maximizing the transfection efficiencies of non-
viral vectors requires a full understanding of the supramolecular structures of CL-DNA
complexes, their interactions with cell membranes and the events leading to release of
DNA inside the cell for possible expression. Recent synchrotron small angle x-ray
diffraction (XRD) work has solved the two types of structures observed in CL-DNA
complexes. These are a multilamellar structure with DNA monolayers sandwiched
between cationic membranes (LαC) [14], and an inverted hexagonal structure with DNA
encapsulated within cationic lipid monolayer tubes (HII
C). Both structures are shown in
Fig. (2) [15]. We are now beginning to understand the parameters governing assembly
and properties of the supramolecular structures of CL-DNA complexes in different lipid-
membrane systems [14,15,19,20,21,22]. The transfection efficiencies of nonviral
delivery methods may be improved through these insights into transfection-related
mechanisms at the molecular and self-assembled levels as we describe below.
Insert Fig. 2
Our work aims at unraveling the mechanisms of transfection by CL-DNA
complexes at the molecular to cellular level. In particular, we focus on the influence of
the structure of the complex and, within a given structure (lamellar LαC or inverted
hexagonal HII
C), the chemical (e.g. membrane charge density) and physical (e.g.
8
mechanical properties of the membrane) parameters of the membranes forming the
complex. As an example, the mechanical properties of a lipid membrane are determined
by its bending rigidity κ and its so-called Gaussian modulus κG. The latter characterizes
the tendency of a membrane to form pores; thus, CL-DNA complexes with large
Gaussian moduli may show enhanced transfection efficiency because of their potential
tendency to form energetically favored pores with endosomal membranes releasing DNA
into the cytoplasm (see e.g. Fig. (10)).
Our studies are carried out by employing synchrotron XRD for structure
determination, reporter gene assays to determine transfection efficiency, and 3-
dimensional laser scanning confocal microscopy (LSCM) to image complex pathways
and their interactions with cells. The combination of these three characterization
methods allows us to interpret the different interactions between high and low
transfection efficiency complexes and cells, as observed in confocal microscopy, based
on knowledge of the structure of the complexes. The experiments utilize commercially
available lipids (cf. Fig. (3)) as well as lipids with multivalent cationic head-groups that
were newly synthesized for the purpose of clarifying the precise role of the membrane
charge density [23,24].
The in vitro studies we describe in this review should apply to TE optimization in
ex vivo cell transfection, where cells are removed and returned to patients after
transfection. In particular, our studies aimed at understanding the key mechanisms
underlying TE in continuous (dividing) cell lines should aid clinical efforts to develop
efficient CL-vector cancer vaccines in ex vivo applications. Those vaccines are intended
to induce transient expression of genes coding for immuno-stimulatory proteins in
9
dividing cells [2,6,10,25,26]. Thus, the nuclear membrane, which dissolves during
mitosis, is not considered a barrier to the delivery and expression of DNA.
Insert Fig. 3
In the final “Future Direction” section of this review we outline ongoing
experiments in our laboratory using a new class of surface-functionalized CL-DNA
complexes, which contain recently synthesized peptide-lipids with specific sequences for
targeting of the cell surface and the nucleus [27]. As we describe, such surface-
functionalized CL-DNA complexes should be suitable for future in vivo and systemic
delivery studies. While the majority of our current effort is focused on optimizing
transfection in continuous cell lines, we point out that experiments on non-dividing cells
blocked at the G1/S point will be feasible with the surface-functionalized CL-DNA
complexes containing peptide-lipids with a nuclear localization sequence (NLS). These
experiments will serve as models of transfection in cell lines that either divide very
slowly or are non-dividing (e.g. differentiated nerve cells).
THE RELATION OF MEMBRANE CHARGE DENSITY AND TRANSFECTION
EFFICIENCY OF CATIONIC LIPID-DNA COMPLEXES
A critical requirement for enhancing the transfection efficiency of synthetic carriers
is a full understanding of the different structures of CL-DNA complexes and the physical
and chemical basis of their interactions with cellular components. To this end, we
examine the structure-dependence of DNA delivery through imaging and transfection
studies of CL-DNA complexes that exhibit one of the two known structural phases, LαC
10
or HII
C [14,15]. Three-dimensional LSCM allows direct imaging of complexes and their
interactions with cells, and reporter gene transfection studies give a statistically
meaningful measure of the total amount of protein synthesized by cells from delivered
DNA.
There are many parameters that can be varied to optimize transfection by CL-DNA
complexes, e.g. lipid structure, lipid charge, lipid to DNA charge ratio, or the amount of
neutral lipid. We are particularly interested in understanding these as factors that
influence the properties of the self-assembled system as a whole, enabling a generalized
view. One of these properties is the membrane charge density σM, defined as the total
charge of the membrane due to the cationic lipid headgroups divided by the total area of
the membrane. For isoelectric CL-DNA complexes in the LαC phase, the average spacing
dDNA between DNA molecules can be expressed in terms of the average distance per
anionic charge along the DNA backbone (l0), and the membrane charge density σM:
dDNA = e/(l0σM) [14,19]. Thus, the DNA interaxial spacing is a measure of σM and
decreases as the membrane charge density increases. For positively charged complexes,
σM can no longer be directly calculated from dDNA, but the same qualitative relationship
still holds.
To check whether σM is a relevant chemical parameter of CL-DNA complexes
relating to TE, we performed experiments in which we varied σM systematically for
lamellar and inverted hexagonal phases. We investigated the effect on TE and also, via
confocal imaging, the pathways of gene delivery into mouse L cells. The structures of
the cationic and neutral lipids used in these first studies are given in Fig. (3). The
cationic lipids were either commercially available or gifts from other investigators. As
11
neutral lipids, we used 1,2-dioleoyl-sn-glycerophosphatidylethanolamine (DOPE) and
1,2-dioleoyl-sn-glycerophosphatidylcholine (DOPC). DOPE is one of the main neutral
lipids currently in use in gene therapy applications of CLs. To fully explore the validity
of our findings, further experiments using new, recently synthesized lipids with
multivalent cationic head groups were performed [23,24].
Prior to our studies, TE measurements by other groups had shown that in mixtures
of the monovalent cationic lipid 2,3-dioleyloxypropyltrimethylammonium chloride
(DOTAP) and neutral lipids, typically at a weight ratio between 1:1 and 1:3, DOPE aided
whereas DOPC severely suppressed transfection [28,29]. Therefore, it seemed that
DOPE-based HII
C complexes always transfect more efficiently than LαC complexes. Our
findings show that the difference in performance between the two types of complexes is
subtler than previously believed and that there are regimes of composition where the TE
of complexes in the LαC phase becomes as high as that of the best HII
C complexes.
Insert Fig. 4
In our first set of experiments, the lipid mixtures for LαC complexes (Fig. (4), left)
were comprised of monovalent cationic DOTAP mixed with neutral DOPC. Those for
HII
C complexes (Fig. (4), right) contained DOTAP mixed with neutral DOPE. We used
positively charged CL-DNA complexes prepared at ρ = DOTAP/DNA (wt./wt.) = 6 with
supercoiled pGL3 (luciferase gene/SV 40 promoter; from Promega Corp.) plasmid DNA.
The isoelectric point, where the charges of the DNA are exactly neutralized by those of
the lipid, is at ρ = 2.2. We chose the weight ratio ρ = 6, equivalent to a cationic to
12
anionic charge ratio of 2.8, because it corresponds to the middle of a typical plateau
region that is observed when plotting TE as a function of increasing ρ above the
isoelectric point. XRD measurements elucidated the structures of these complexes in
water and in Dulbecco’s Modified Eagle Medium (DMEM), a common environment for
in vitro studies of cells [24,38,30]. Synchrotron small angle x-ray scattering (SAXS)
patterns of DOTAP/DOPC complexes at the mole fraction ΦDOPC = 0.67 (Fig. (4), left)
showed sharp peaks at q001 = 0.083 Å-1, q002 = 0.166 Å-1, with a shoulder peak at
q003 = 0.243 Å-1 and q004 = 0.335 Å-1, resulting from the layered structure of the LαC phase.
The interlayer spacing is d = δm+ δw = 2π/q001 = 75.70 Å, with DNA intercalated between
cationic lipid bilayers (cf. Fig. (2)). For DOTAP/DOPE complexes at ΦDOPE = 0.69,
SAXS (Fig. (4), right) revealed four orders of Bragg peaks at q10 = 0.103 Å-1, q11 = 0.178
Å -1, q20 = 0.205 Å-1 and q21 = 0.270 Å-1, denoting the HIIC phase with a unit cell spacing a =
4π/[(3)1/2q10] = 70.44 Å (Fig. (4), right, inset and Fig. (2)). Mouse L cells were
transfected with CL-DNA complexes using pGL3 DNA that contains the firefly
luciferase reporter gene. The TE was measured via the standard luminescence assay as
relative light units (RLU) per mg of cell protein [38,30]. The results of these
experiments are shown in Fig. (4) (middle), demonstrating that the TE attainable with
complexes in the HIIC phase at ΦDOPE = 0.69 is by more than two decades higher than that
of LαC complexes at ΦDOPC = 0.67.
Insert Fig. 5
13
To further understand the structure-function correlation, we examined the transfer
process of CL-DNA complexes into cells and the mechanism of the subsequent DNA
release using LSCM, which provides an optical resolution of ~ 0.3 µm in the x and y, and
~ 3/4 µm in the z direction. By comparing images of the x-y, y-z, and x-z planes, it is
possible to determine the position of an object relative to a cell. The complexes were
doubly tagged with fluorescent labels, Texas Red-DOPE for lipid and a covalently
attached green label for DNA (Mirus Label ITTM) [38]. The intensity of the fluorescence
sets the detection limit for DNA and lipid, which does not reach the single molecule level
for uncondensed DNA. Fig. (5) shows LSCM micrographs of mouse L cells taken six
hours after the addition of complexes. In Fig. (5 (A)), a typical image of a cell
transfected with HIIC complexes at ΦDOPE = 0.69 is displayed. The lipid fluorescence
clearly outlines the plasma membrane, indicating either spontaneous transfer of labeled
lipid or fusion of lipid with the plasma membrane (which may have occurred before or
after entry through the endocytic pathway [31]). An aggregate of complexes (yellow) is
visible inside the cell as well as lipid-free DNA (green) in the cytoplasm. The image
shows that the interaction between HII
C complexes and cells leads to the dissociation and
release of DNA from the CL-vector, consistent with the measured high TE.
The corresponding confocal images for LαC complexes at ΦDOPC = 0.67 are shown in
Fig. (5 (B)). In striking contrast to HIIC complexes, we observed no free DNA. Rather,
many individual intact CL-DNA complexes are found inside of the cells. A typical
complex is highlighted in Fig. (5 (B)). These results show that at ΦDOPC = 0.67, most of
the DNA remains trapped by the CL-vector, consistent with the measured low TE. As
there is no fluorescent lipid observed in the cell membrane, fusion is ruled out as a
mechanism of cell entry. Thus, the complexes most likely entered the cells through
14
endocytosis, as also observed by others [31,32,33]. Furthermore, transfection
experiments performed in the presence of chloroquine showed that the intact CL-DNA
complexes were mostly trapped within endosomes as we will discuss later.
Insert Fig. 6
In further experiments, we observed an unexpected enhancement of the TE by two
decades as the concentration of DOPC in LαC CL-DNA complexes was reduced. In
Fig. (6 (A)) (diamonds), the non-trivial dependence of TE on ΦDOPC for DOPC/DOTAP-
DNA complexes is plotted. The transfection efficiency starts low for 0.5 < ΦDOPC < 0.7
and increases dramatically to a value, at ΦDOPC = 0.2, which rivals that achieved by the
DOPE/DOTAP-DNA HII
C complexes. Similar results were obtained for another
univalent cationic lipid 2,3-di(myristyloxy)propyl(2-hydroxyethyl)dimethylammonium
bromide (DMRIE) (Fig. (6 (A)), triangles). The key experiment, which led to a deeper
understanding of the observed TEs, was a study done with the multivalent cationic lipid
2,3-Dioleyloxy-N-[2-(sperminecarboxamido)ethyl]-N,N-dimethyl-1-propylammonium
chloride (DOSPA) (Fig. (6 (A)), squares) replacing DOTAP. A qualitatively similar
trend was observed, with TE decreasing rapidly above a critical Φ∗DOPC. However, Φ∗
DOPC
was shifted from ~ 0.2, as observed for DOTAP and DMRIE complexes, to 0.7 ± 0.1 for
DOSPA. The main difference between the cationic lipids is their headgroups (Fig. (6),
inset). The headgroup of DOSPA has a high charge of up to +5 at a size that is only
somewhat larger than that of the DOTAP headgroup. Thus, the charge density (charge
per unit area) of the headgroup is much higher for DOSPA and, at a given ΦDOPC, the
15
membrane charge density σM in complexes containing DOSPA also is significantly larger
compared to complexes prepared from DOTAP or DMRIE.
In Fig. (6 (B)), the same TE data as in Fig. (6 (A)) is shown, now plotted versus the
average membrane charge density σΜ:
σΜ = eZNcl/(NnlAnl + NclAcl) = [1- Φnl/(Φnl + rΦcl)]σcl (1)
In equation (1) we have defined r = Acl/Anl as the ratio of the head-group area of the
cationic to neutral lipid, σcl = eZ/Acl as the charge density of the cationic lipid with
valence Z, and Φnl = Nnl/(Nnl + Ncl) and Φcl = Ncl/(Nnl + Ncl) as the mole fractions of the
neutral and cationic lipids, respectively. For the plots in Fig. (6 (B)), we used Anl = 72 Å2
[34], rDOTAP = rDMRIE = 1, rDOSPA = 2, ZDOTAP = ZDMRIE = 1, ZDOSPA = 4 (expected for pH ≈ 7).
Since we were unable to obtain well-defined values for the head group area of cationic
lipids using standard methods, r is an adjustable parameter. The values used in the plots
agree well with chemical intuition (cf. also inset in Fig. (6)). All other parameters are
fixed by the experiment or the chemical structure. Given the complexity of the CL-
DNA-cell system, it is remarkable that the data, spread-out when plotted as a function of
Φnl (Fig. (6 (A))), coalesce into a single, “universal” curve as a function of σΜ, with TE
varying exponentially over nearly four decades as σΜ increases by a factor of ≈ 8
(Fig. (6 (B)), σΜ between 0.0015 e/Å2 and 0.012 e/Å2). This clearly implies that σΜ is a
key universal parameter for transfection with lamellar LαC CL-vectors. We now observe
a single optimal σM* ≈ 0.0104 ± 0.0017 e/ Å2 ≈ e/(100 Å2) (Fig. (6 (B)), arrow) beyond
which (σΜ > σM*) the universal TE curve saturates for both univalent and multivalent
16
cationic lipid containing CL-vectors. The membrane charge density σΜ controls the
average DNA interaxial spacing dDNA (cf. Fig. (2)) [14,19], which was found to decrease
as σΜ increases and level off for σΜ > σM* [38]. It is important to note that all TE
measurements were done with 2 µg of plasmid DNA at a constant cationic to anionic
charge ratio of 2.8 (see above). Thus, every TE data point used the same amount of total
charged species (anionic charge from DNA, cationic charge from cationic lipid) and only
the membrane charge density of CL-DNA complexes (σΜ) was varied by adjusting the
amount of neutral lipid.
Insert Fig. 7
The TE data show vastly diverse behaviors of LαC CL-DNA complexes between
low and high σΜ. As discussed earlier, for low σΜ = e/(200 Å2), corresponding to low
TE, confocal images show DNA locked within complexes after endocytosis
(Fig. (5 (B))). To test the idea that the complexes are trapped in the endosome, we
carried out transfection experiments in the presence of chloroquine. This is a well-
established bioassay known to enhance the release of material trapped within endosomes
[35] by osmotically bursting late-stage endosomes. The endocytic pathway involves
lowering the pH inside the endosome and fusion of the endosome with nuclease
containing lysosomes, leading to late-stage endosomes. This limits the time available for
CL-DNA complexes to escape. Chloroquine is a weak organic base, which can penetrate
lipid membranes in the non-protonated state. Once protonated, it can no longer diffuse
through the membrane and therefore effectively acts as a buffer selective for acidic cell
17
compartments. As more and more protons are pumped into the buffered endosome to
lower the pH, counterions follow, which eventually causes rupture of the vesicle due to
the increased osmotic pressure [36]. The fractional increase (TEchloroquine/TE; note the
logarithmic scale) for the DOSPA/DOPC and DOTAP/DOPC systems with added
chloroquine, plotted as a function of σM (Fig. (7)), shows a large increase by as much as a
factor of 60 as σΜ decreases. This indicates that at low σΜ, lamellar LαC complexes are
trapped within endosomes, consistent both with the confocal images (Fig. (5 (B))) and
the measured low TE without chloroquine. At high σM, chloroquine has a much smaller
effect on TE with the fractional increase of order unity, implying that endosomal
entrapment is not a significant limiting factor.
A comparison of TE as a function of σΜ for DOTAP/DOPC-DNA and
DOTAP/DOPE-DNA complexes is shown in Fig. (8). The DOTAP/DOPE system goes
through 2 phase transitions: LαC (filled squares) to coexisting Lα
C + HII
C (squares with
cross) to HII
C (open squares). At high ΦDOPE > 0.56, DOTAP/DOPE-DNA complexes are
in the HII
C phase (open squares) and exhibit high TE. In contrast to DOPC containing
complexes, which show a strong dependence on the mole fraction of neutral lipid and
therefore σΜ, the TE of DOPE-containing complexes is independent of σΜ. In contrast,
high TE requires σΜ > σM* for LαC complexes. We can thus conclude from the data that
σΜ is an essential parameter for transfection with LαC complexes but not HII
C complexes.
The mechanism of transfection by DOPE containing HII
C complexes is dominated by
other effects as we describe below.
Insert Fig. 8
18
LSCM images of cells transfected with LαC complexes at high σΜ displayed a path
of complex uptake and DNA release distinct from both LαC complexes at low σΜ (Fig. (5
(B))) and HII
C complexes (Fig. (5 (A))). A typical confocal image of a cell incubated for
6 h with LαC complexes at ΦDOPC = 0.18 (σΜ ≈ 0.012 e/Å2) is shown in Fig. (9). A few
intact complexes are visible inside the cell: Fig. (9) (label 2; box 2) shows the equal
green (DNA) and red (lipid) fluorescence intensity along the dotted line in the inset.
More interestingly, a mass of exogenous DNA successfully transferred into the
cytoplasm was also clearly evident (Fig. (9), label 1; box 1 shows the much larger green
(DNA) fluorescence intensity along the x-y diagonal). Since there is no indication of
transfer of fluorescent lipid to the plasma membrane, the complexes with high σΜ must
have entered the cells through endocytosis. The integrated fluorescence intensity of the
observed DNA (Fig. (9), box 1) is comparable to that of DNA complexed with lipids
(Fig. (9), box 2), indicating that the released DNA is in the form of aggregates. As
endosomes contain no known DNA-condensing agent, these aggregates must reside in
the cytoplasm, which contains many multivalent cationic biomolecules (e.g. spermine
and histones, which become available during the cell cycle) that are able to condense
DNA [37]. The presence of lipid-released DNA in the cytoplasm after endocytic uptake
of complexes is in agreement with the measured high TE and, moreover, implies fusion
between CL-DNA lipids and endosomal membranes, enabling escape from the
endosome. This is consistent with our finding that chloroquine has a small effect on TE
at high σΜ, suggesting that endosomal escape is not a major obstacle. The confocal
image also shows a large aggregate of complexes in on part of the cell (Fig. (9), label 3).
19
Again, there is no transfer of fluorescent lipid to the cell membrane, which rules out
entry by fusion with the plasma membrane. Comparing the changes in fluorescence
intensity along the x-y-diagonal (Fig. (9), box 3) and z-axis (Fig. (9), box 4), from the
outside toward the inside of the cell, we see an aggregate of complexes caught in the
process of dissociation after endocytosis, with released DNA toward the inside of the
cell.
Insert Fig. 9
In summary, our 3-dimensional LSCM studies have revealed distinct interactions
between CL-DNA complexes and mouse L cells, both for lamellar LαC and inverted
hexagonal HII
C nanostructures. Confocal images of LαC complexes in cells identified two
regimes [38]. For low σM, DNA remained trapped in CL-vectors. By contrast, for high
σΜ, released DNA was observed in the cytoplasm, indicative of escape from endosomes
through fusion. TE studies using the firefly luciferase reporter gene revealed a truly
unexpected result: at a constant cationic to anionic charge ratio, TE data for univalent and
multivalent cationic lipids merged into a single curve as a function of σM, identifying it as
a key universal parameter [38]. The universal TE curve revealed an optimal membrane
charge density (σM*) where LαC complexes with σM > σM* achieve TEs comparable to that
of highly transfecting HII
C complexes. In contrast to HII
C complexes, where TE is
independent of σM, the universal TE curve for LαC complexes was found to increase
exponentially over nearly four decades with increasing σM < σM*.
20
A model of cell entry by LαC CL vectors: Dependence on σΜ and elastic moduli
of cationic membranes of CL-DNA complexes. The combined x-ray diffraction,
LSCM, and TE data lead to a model of cellular entry via LαC CL carriers (Fig. (10)).
Previous work indicates that the electrostatic attraction between cationic CL-DNA
complexes and mammalian cells is mediated by negatively charged cell surface sulfated
proteoglycans (Fig. (10), label a) [39]. Our LSCM images show no evidence of fusion
with the plasma membrane, implying that LαC complexes enter cells via the endocytic
pathway (Fig. (10), labels b and c). This is consistent with results of other research
groups and our transfection experiments in the presence of chloroquine. Once inside the
cell, LαC CL-DNA particles exhibit two distinct types of behavior as revealed by confocal
microscopy. At low σΜ, mostly intact LαC complexes are observed inside of the cells and
TE experiments in the presence of chloroquine are consistent with intact complexes
trapped in endosomes (Fig. (10), label c). At high σΜ, LSCM revealed lipid-free DNA
inside cells. Since the observed DNA is in a condensed state, it must reside in the
cytoplasm (Fig. (10), label e) because of the lack of condensing molecules in the
endosome [37]. This observation is consistent with TE data at high σΜ, where the
addition of chloroquine does not increase TE.
Insert Fig. 10
From all our data, we conclude that in the regime where transfection efficiency
scales exponentially with σΜ, the limiting step in the transfection process is escape from
the endosome after endocytosis. The most probable mechanism for this is through fusion
21
with the endosomal membrane (Fig. (10), label d). The model we propose connects this
process with the membrane charge density σΜ of enclosed complexes. For σΜ < σM*,
the universal TE curve (Fig. (6 (B))) shows TE increasing exponentially as a function of
σΜ and therefore suggests a simple model of activated fusion (showing Arrhenius
behavior) of complexes with endosomal vesicles. Since fusion is the rate limiting
process, TE is proportional to the rate of fusion:
Transfection Efficiency∝ rate of fusion = τ -1 [exp(-δE/kBT)], (2)
with δE = energy barrier height = a·κ – b·σM . Here, a, b > 0 are constants and τ -1 is
the collision rate between the trapped CL-DNA particle and the endosomal wall (Fig.
(10), label c). The membrane bending modulus κ is a measure of the flexibility of the
membrane [40,41,42]. The larger κ, the higher the energy cost associated with bending
the membrane. Bending is inevitable in the fusion of membranes, as shown schematically
by the arrows in the expanded view of Fig. (10), label d. The energy required for this,
which is proportional to κ, thus provides the main barrier to fusion. On the other hand,
electrostatic attraction favors membrane adhesion because the fusing membranes are
oppositely charged. This results in a lower energy barrier as σΜ increases (Fig. (10),
label c, arrows in expanded view), corresponding to the term “– b·σM ” in equation (2).
The fact that higher σΜ lowers the barrier height and thus appears in the exponent in
equation (2) is responsible for the exponential increase of TE observed in our
experiments (Fig. (6 (B))). In agreement with our model, a recent theoretical study,
which considers fusion between two neutral lipid bilayers (i.e. forming pores between
22
neighboring membranes), has found that the main energy barrier against fusion is
proportional to κ [43]. The same study also implies that the membrane Gaussian
modulus κG should enhance fusion when κG > 0, favoring pore formation. Current work
in our group, e.g. with lipids showing a preference for cubic phases (with κG > 0), is
aimed at experimentally testing the influence of κG and κ on TE.
A model of cell entry by CL-vectors with the HII
C structure: Relevance of the
outermost lipid layer. A striking difference between transfection by HII
C versus LαC
complexes is that for the former, transfection efficiency is independent of the membrane
charge density σM (Fig. (8)). Thus, a different mechanism, which is independent of σΜ,
must dominate the interaction between HII
C complexes and cells. The lipid fluorescence,
which outlines the plasma membrane in the LSCM images shown in Fig. (5 (A)),
indicates mixing of lipids of HII
C complexes with the plasma membrane, which may have
occurred before or after entry through the endocytic pathway. The image shows that the
interaction between HII
C complexes and cells leads to dissociation and release of DNA
from the CL-vector consistent with the measured high TE. A mechanism that may be
responsible for rapid fusion is shown schematically in Fig. (11). In the top part, a HII
C
complex is shown approaching either the plasma or the endosomal membrane. The cell-
surface proteoglycans, which again mediate the attraction between the complex and the
membrane (cf. Fig. (10)), have been omitted for the sake of clarity. For lipids forming
HII
C complexes, the preferred membrane curvature is negative, as opposed to ≈ 0 for
lipids that form the LαC structure. Negative curvature is realized for the lipids coating
DNA inside the HII
C complex, but the curvature of the outermost lipid monolayer, which
23
must cover the HII
C complex to provide a hydrophilic surface, is positive. This elastically
frustrated state of the outer monolayer, which is independent of σM, drives the rapid
fusion with the plasma or endosomal membrane, leading to release of a layer of DNA
and a smaller HII
C complex as shown in the bottom part of the Fig. (11). In the language
of our model for cell entry of LαC complexes, the activation energy for fusion of HII
C
complexes with the endosomal membrane is negligible and this step no longer limiting.
The outer layer of the released, smaller HII
C complex is again elastically frustrated and
will drive quick release of the remaining DNA through interactions with other
membranes, negatively charged cellular proteins or DNA-condensing molecules. By
comparison, the bilayers of lamellar LαC complexes are inherently more stable. After
escaping the endosome through fusion (see Fig. (10), label d), the onion-like (lipid-
bilayer/DNA-monolayer) complex is expected to peel much more slowly, layer-by-layer,
through interactions of the cationic membranes with anionic components of the cell, such
as the predominantly anionic cytoskeletal filaments [44], or DNA-condensing
biomolecules.
Insert Fig. 11
Structures and transfection properties of new, ornithine-based multivalent
cationic lipids. To further investigate our finding of a well-defined relationship between
transfection efficiency and membrane charge density σM in lamellar LαC complexes, and
to explore more broadly the relevance of σM as a key chemical parameter, we have
recently synthesized new lipids with multivalent cationic head groups, which we describe
24
in this section. The general structure of the lipids is shown in Fig. (12), and the lipid
head groups are compiled in Table 2. The headgroups are based on the amino acid
ornithine, whose structure is shown in Table 2 in the top-left box. A key point is that
their charge was varied systematically from +2 (MVL2) to +5 (MVL5). This was
achieved by the addition of propyl-amine groups to the original ornithine via Michael
addition of acrylonitrile and subsequent hydrogenation [24]. A monovalent lipid
(MVL1), based on glycine, was also prepared for comparison.
Insert Fig. 12
Insert Table 2.
Work on one of the new multivalent cationic lipids (MVL5) shows structurally
stable complexes with improved transfection efficiency behavior [24]. Optical
micrographs of these CL-DNA complexes are shown in Fig. (13), imaged in differential-
interference-contrast (left), DNA fluorescence (middle) and lipid fluorescence mode
(right). The complexes were prepared using DOPC/MVL5 lipid mixtures which
contained 40 wt.-% MVL5 at a cationic lipid to DNA charge ratio of 2.8, which gives
optimal transfection for complexes made from DOTAP and DOPC [38]. The
observation of co-localization of lipid and DNA by fluorescence microscopy proves the
formation of complexes.
Insert Fig. 13
25
The structure of the CL-DNA complexes was investigated by XRD. In Fig. (14),
SAXS patterns are shown for complexes at weight fractions of cationic lipid of 40%
(MVL5) and 30% (DOTAP), corresponding to equivalent molar fractions of 31% and
33% of cationic lipid in the membrane. At this composition, the transfection efficiency
of the MVL5 complexes is approximately 100-fold higher than that of the DOTAP
complexes. The lipid to DNA charge ratio for all complexes was again 2.8. Both CL-
DNA complexes are in the lamellar LαC phase. As with DOTAP, the Lα
C phase is
observed throughout the lipid composition range for MVL5/DOPC/DNA complexes.
For the MVL5 complexes in Fig. (14), three peaks at q = 0.088 Å-1, 0.175 Å-1 and 0.261
Å -1 are observed, corresponding to the (00h) peaks of the layered structure with an
interlayer spacing δ = 2π/q001 of 71.2 Å.
Insert Fig. 14
The broad peak at qDNA = 0.201 Å-1 arises from correlations between DNA chains
within a water gap and gives their interaxial spacing dDNA = 2π/qDNA = 31.2 Å [14,15].
This spacing increases (qDNA = 0.106 Å-1, i.e. dDNA = 59.5 Å) for the complexes prepared
with an equivalent molar ratio of the univalent lipid DOTAP, reflecting the lower
membrane charge density in DOTAP/DOPC complexes due to the smaller head group
charge.
Insert Fig. 15
26
For transfection studies, we compared CL-DNA complexes prepared with DOTAP
and MVL5. While the ratio of neutral to cationic lipid was varied, all complexes again
had the lipid to DNA charge ratio of 2.8 optimized for the DOTAP/DOPC system. By
assigning a charge of +4 to it, incomplete protonation of the MVL5 head group at neutral
pH [45] was taken into account. Transfection experiments using mouse fibroblast L-cells
and a luciferase reporter assay are displayed in Fig. (15). The new lipid MVL5 gives
higher transfection efficiencies for all ratios of neutral to cationic lipid. However, the
difference in transfection efficiencies increases dramatically from one to three orders of
magnitude as the amount of cationic lipid is reduced from 50 to 20 mol-percent. For
MVL5, the transfection efficiency exhibits a broad maximum and remains high, while it
drops quickly for DOTAP. This again demonstrates the importance of the membrane
charge density for transfection with LαC complexes as described in the preceding section
[38].
FUTURE DIRECTIONS
Thus far we have considered CL-DNA complexes whose interactions with cells are
non-specific and typically mediated by electrostatic forces. In this section we describe
ongoing experiments where directed interactions (i.e. ligand-receptor type) are added to
CL-DNA complexes by incorporating peptide-lipids with specific amino-acid sequences.
We have recently synthesized such lipids, which contain peptides on hydrophilic
poly(ethylene glycol) (PEG) spacers of varied length. The general structure of the
peptide-lipids is shown in Fig. (16) and examples of employed peptide ligands and
spacers are given in Table 3. We are pursuing two major directions in this research
effort. One are CL-DNA complexes that will target the nucleus via a nuclear localization
27
sequence. The other are complexes suitable for future systemic or in vivo studies, which
will target cell surface receptors [27].
Insert Fig. 16
Insert Table 3.
Surface Functionalized CL-DNA Complexes containing NLS Peptide-Lipids:
CL-DNA Complexes for Transfecting Non-Dividing Cells. A major continuing focus
of work in our laboratory, and indeed for a large number of other laboratories worldwide
working on non-viral gene delivery systems, is on optimizing transfection at the cell
level in continuous (dividing) cell lines with potential relevance for the development of
cancer vaccines [1,2,3,4,5,6,7]. Indeed, a large fraction of clinical work using CLs for
delivery is centered around CL-DNA vaccines intended to induce transient expression of
genes encoding for immuno-stimulatory proteins in dividing cells. The nuclear
membrane, which dissolves during mitosis, is not a barrier in these experiments.
However, even cancer cells are dividing much more slowly than the cells used for
in vitro experiments and some very interesting target cells for gene delivery, such as
nerve cells, are non-dividing cells. Thus, vectors that are able to overcome the barrier
posed by the nuclear membrane are of great interest. In the case of lamellar complexes,
confocal microscopy shows that while released DNA is only at high membrane charge
density, intact complex particles are present in the cell for both high and low membrane
charge density. (Fig. (5 (B)) and Fig. (9)). In addition, our proposed mechanism entails
the release of a smaller complex particle into the cytoplasm after fusion with the
endosomal membrane. Thus, intracellular targeting of a fraction of the complexes via
28
peptide lipids should be feasible. We are investigating whether peptide-lipids containing
the NLS of the SV 40 virus [46] can mediate transport of the fraction of CL-DNA
complexes with a particle size of less than 25 nanometers through the nuclear pore.
Colloidal particles of up to 25 nm diameter, when coated with SV 40 NLS peptide-
conjugated bovine serum albumin (BSA), are transported into the nuclei of HeLa cells
[47,48]. While dynamic light scattering data gives a typical average CL-DNA complex
size of around 100 to 150 nm, the distribution of sizes is very broad, extending down to
10s of nanometers [4,14,15]. The larger particles observed in microscopy are likely
aggregates of smaller particles. Furthermore, cryo electron microscopy data has shown
that a fraction of complexes consists of as little as 2 to 3 lipid-bilayers complexed with
DNA [49]. Optimizing TE for complexes that can cross the nuclear membrane is
important for future delivery applications in slowly or non-dividing cells. Once inside
the nucleus, the DNA will rapidly associate with histones or polyamines. We have
observed both these processes in vitro (H. M. Evans, C. R. Safinya, unpublished and
[21]).
Surface-functionalized CL-DNA complexes containing RGD-based peptide-
lipids: CL-DNA complexes for in vivo applications. The attachment of CL-DNA
complexes to mammalian cells and the resulting interactions (cf. Fig. (10), label a) are
mediated through electrostatic attractions between cationic CL-DNA complexes and
negatively charged cell surface sulfated proteoglycans [39]. We are investigating an
RGD-peptide with the sequence GRGDSP, attached to a lipid to investigate alternative
pathways of cell adhesion. The RGD sequence is expected to bind with high-affinity to
the cell’s surface integrins (i.e. αvβ3 and αvβ5), leading to receptor-mediated endocytosis.
A main motivation for these experiments is that the concept could later be extended to
29
more specific ligands with the aim of targeting specific cell types for systemic in vivo
applications.
Insert Fig. 17
CL-DNA complexes, as they are currently used in ex vivo and in vivo clinical
trials, the latter involving intra-tumoral injection methods [1,2], are not suitable for
systemic in vivo applications. Cationic lipids and their complexes with DNA activate the
complement system [50], resulting in their rapid removal from circulation by the
mononuclear phagocytic system cells through the process of opsonization. PEG-
conjugation to poly(L-lysine) can reduce the activation of the complement system [50], a
phenomenon well known for liposomes [51,52]. The presence of a hydrophilic
polymeric shell on liposomes provides a repulsive barrier and results in vastly increased
circulation lifetimes, which is referred to as steric stabilization. Thus, complexes
designed for systemic delivery should contain a PEG coating to prevent opsonization.
However, once the PEG layer is thick enough to effectively minimize undesired
interactions, the complex will also no longer attach to cells and attachment through a
specific ligand (with RGD serving as a model system) will be needed. The feasibilty of
this concept has been demonstrated for liposomes with antibodies as targeting moieties
[53].
We have performed a series of transfection and XRD experiments using PEG-lipids
of various lengths. We refer to these as PEGMW–lipids, where MW is the molecular
weight of the PEG chain. Recent x-ray data show that CL-DNA complexes containing
PEG2000-lipid, shown schematically in the top part of Fig. (17), are indeed structurally
30
stable and thus suitable candidates for the incorporation of additional RGD-PEG-lipids
[54]. To demonstrate that the PEG2000-lipid is also exposed at the complex surface, we
conducted TE measurements of highly transfecting CL-DNA complexes (80 mole%
DOTAP) as a function of increasing amount of incorporated PEG-lipid (Fig. (17),
bottom). The addition of about 6 mol-% PEG2000-lipid already leads to a strong
suppression of TE, by about 2 orders of magnitude. This indicates that the electrostatic
binding of the cationic CL-DNA complex to cells is efficiently reduced due to the PEG2000
polymer coat of thickness ≈ 35 Å [31]. In contrast, the shorter PEG chain of a PEG400-
lipid does not provide significant shielding of the electrostatic interactions and TE
remains relatively high. Note the difference in mol-fraction for the PEG-lipids, which
correspond to approximately equal total amounts of PEG.
We are now investigating surface-functionalized CL-DNA complexes, which
incorporate a mixture of PEG2000-lipids and RGD-PEG2000-lipids within their membranes.
The RGD-peptide should provide an effective ligand for the integrin receptors on the cell
surface. For such functionalized CL-DNA complexes, the primary role of the cationic
lipid is the condensation of DNA. This is in contrast to standard CL-DNA complexes,
where the cationic lipids both condense DNA and provide for electrostatic attachment to
the cell surface. Therefore, the new complexes are not only designed to be suitable for
systemic in vivo applications due to their steric stabilization, but the reduced amount of
cationic lipid will also reduce their toxicity.
CONCLUSIONS
The broad, long-term objective of our research is to develop a fundamental science
base, which will lead to the design and synthesis of optimal nonviral carriers of DNA for
31
gene therapy. Simultaneously, a major long term objective is to improve efficiency for
delivering large pieces of DNA containing important human genes and their regulatory
sequences (> 100 k-base-pairs), which at present can only be achieved with synthetic
vectors. The structure-function data obtained from our research should allow us to begin
the formidable task of rationally designing these self assemblies for enhanced gene
delivery, beginning with the chemical structure of the lipids and the appropriate lipid
compositions.
Our work has shown that CL-DNA complexes of distinct structure (i.e. LαC versus
HII
C) differ widely in their interactions with cells and their ability to deliver exogenous
DNA to the cytoplasm. We have demonstrated that σM, the average membrane charge
density of the CL-vector, is a key universal parameter that governs the transfection
behavior of LαC complexes in cells. The universal TE curve for DOPC-containing Lα
C
complexes increases exponentially with σΜ for σΜ < σΜ* (an optimal membrane charge
density), and saturates for σΜ > σΜ*. The limiting saturated TE level of LαC complexes
is comparable to the high TE of DOPE-containing HII
C complexes, which exhibit no
dependence on σΜ. For both LαC and HII
C complexes, confocal microscopy reveals that
much of the lipid-released DNA is in a condensed state and therefore has to contain
oppositely charged macro-ion condensing agents from the cytoplasm. While DNA
strands covering the surface of the DNA mass or isolated from it are most likely
transcriptionally active, much of the observed bulk of condensed DNA may be
transcriptionally inactive, probably setting the current limit to transfection by cationic
lipid vectors. Future studies should reveal well-defined structure-function correlations
for transfection in vivo, in particular, for local intra-tumoral injection in clinical studies.
32
ACKNOWLEDGEMENTS
We gratefully acknowledge the contributions of recent collaborators, in particular,
Joachim Rädler, Ilya Koltover, Cyril George, Tim Salditt, Uwe Schulze, and Hans-
Werner Schmidt. We have benefited over the years through extensive discussions with
Phillip Felgner, Leaf Huang, Robijn Bruinsma, Tom Lubensky, Philip Pincus, Bill
Gelbart, and Avinoam Ben-Shaul. The work was supported by grant GM-59288 from
the National Institute of General Medical Sciences of the National Institute of Health on
DNA-lipid gene delivery studies. Support was also provided by the National Science
Foundation DMR 0203755, and National Institute of Health grants AI-20611 and AI-
12520. DOSPA and DMRIE were gifts from Phillip Felgner for which we are grateful.
The synchrotron x-ray diffraction experiments were carried out at the Stanford
Synchrotron Radiation Laboratory, which is supported by the U.S. Department of
Energy. The Materials Research Laboratory at UC Santa Barbara is supported by NSF-
DMR-0080034.
33
Tables
Table 1. Open gene therapy clinical trials worldwide as of May 2003.
Number of open clinical
trials worldwideDiseases treated
9LUDO�9HFWRUV
Retrovirus 161 Cancers, SCID, AIDS, hemophilia,others
Adenovirus 135 Cancers, cystic fibrosis, peripheralartery disease, others
Pox virus 37 CancersAdeno-associated virus 9 Cystic fibrosis, hemophilia, prostate
cancerOther viral 6 Cancers
1RQYLUDO�9HFWRUV
Lipofection 57 Cancers, cystic fibrosis, coronaryartery disease, restenosis
Naked DNA 59 Cancers, artery diseases, othersRNA transfer 6 CancersGene gun 4 Skin cancers
34
Table 2. Newly synthesized multivalent cationic lipids. A spacer n of 2, 8, 22 and 43
corresponds to a PEG molecular weight of 150, 400, 1000 and 2000, respectively.
Head group Spacer (EO)n Maximum headgroup charge
Lipid name
NH3+
NH3+
O
02, 8, 22, 43 (ester linkage)
+2 MVL2
NH2+
NH3+
O
NH3+
0 +3 MVL3
NH2+
NH2
+
O
NH3+
NH3+
0 +4 MVL4
NH2+
NH
+
O
NH3+
NH3+
NH3+
0, 2 +5 MVL5 / TMVL5 (n=2)
NH3+
O
0 +1 MVL1
35
Table 3. Compilation of recently synthesized peptide-lipids.
Peptide Sequence Spacer
GPKKKRKV(SV 40 NLS)
PEG 400 (n =9)PEG 1000 (n=23)
GRGDSP PEG 2000 (n = 43)
36
Figure Captions
Fig. 1
Schematic of CL-DNA complex formation: (A): Cationic liposomes, closed spherical
shells of bilayers consisting of cationic and neutral lipids; (B): DNA, with cationic
counterions condensed on the backbone. Scale bar for (A) only. Entropy gain from
release of tightly bound counterions (C) is the main driving force for complex formation.
(D) A typical optical micrograph (differential interference contrast mode) of CL-DNA-
complexes.
Fig. 2
Mixing of DNA and CLs results in the spontaneous formation of CL-DNA complexes
with well-defined self-assembled structures. The schematic on the left shows the local
structure of their interior on the nanometer scale for the most commonly observed
lamellar phase (denoted LαC). As derived from synchrotron x-ray diffraction data, the
structure consists of alternating lipid bilayers DNA monolayers. The interlayer spacing is
d = δw + δm. For certain lipids, mixing of DNA and CLs results in the spontaneous
formation of a different self-assembled structure (denoted HII
C) of CL-DNA complexes, as
shown on the right. The internal structure of the inverted hexagonal HII
C phase of CL-
DNA complexes is comprised of DNA coated with a lipid monolayer arranged on a
hexagonal lattice, as derived from synchrotron x-ray diffraction data. Neutral and
cationic lipids are depicted as having white and gray headgroups, respectively. Adapted
from [14,15].
37
Fig. 3
Chemical structures of cationic and neutral lipids with their abbreviated names.
Fig. 4
Comparison of CL-DNA complex structure and transfection efficiency. Complexes were
prepared using pGL3 plasmid DNA. Left: Small angle x-ray scattering pattern and
schematic view (inset) of lamellar LαC complexes (mole fraction ΦDOPC = 0.67). Right:
Small angle x-ray scattering pattern and schematic view (inset) of inverted hexagonal
(inset) HII
C complexes (mole fraction ΦDOPE = 0.69). Middle: TE of the complexe s, as
measured by luciferase enzyme assays of transfected mouse L cells. Adapted from [38].
Fig. 5
Laser scanning confocal microscopy images of transfected mouse L cells fixed six hours
after incubation with complexes. Red and green fluorescence corresponds to lipid and
DNA labels, respectively; yellow, the overlap of the two, denotes CL-DNA complexes.
For each set, middle is the x-y (top) view at a given z; right is the y-z side view along the
vertical dotted line; bottom is the x-z side view along the horizontal dotted line. Arrows
in the x-z and y-z plane side views indicate objects in circles. (A): Cells transfected with
HII
C complexes (ΦDOPE = 0.69), show fusion of lipid with the cell plasma membrane and the
release of DNA (green in the circle) within the cell. Thus, HII
C complexes display clear
evidence of separation of lipid and DNA, which is consistent with the high transfection
efficiency of such complexes. Very significantly, the observed DNA is in an aggregated
state, which implies that it resides in the cytoplasm and has escaped the endosome. (B):
38
Cells transfected with LαC complexes at ΦDOPC = 0.67 which results in a low membrane
charge density σΜ • 0.005 e/Å2 and low transfection efficiency as shown in Fig. (4). No
fusion is visible and intact CL-DNA complexes such as the one marked with the circle are
observed inside the cells. The observation of intact complexes implies that DNA remains
trapped within the complexes, which is consistent with the observed low transfection
efficiency. The cell outline was observed in reflection mode and therefore appears as
blue. Bars = 5 µm for all planes. Adapted from [38].
Fig. 6
(A) Transfection efficiency as RLU per mg total cellular protein, plotted as a function of
varying mole fraction DOPC with cationic lipids DOSPA, DOTAP, and DMRIE. (B) TE
plotted versus the membrane charge density σM, demonstrating universal behavior of
complexes containing cationic lipids with different charge and head group area (inset). For
all three systems, TE increases with σM until an optimal σM * (arrow at • 0.0104 e/Å2;
determined by the intersection of two straight lines fit to the data (dashed lines) above and
below the “knee”), at which TE plateaus. Adapted from [38].
Fig. 7
The fractional TE increase TEchloroquine/TE for the DOSPA/DOPC and DOTAP/DOPC systems
with added chloroquine as a function of σM. Note the logarithmic scale. The fractional
increase is substantial for low membrane charge densities, indicating that endosomal escape
is limiting in this regime. Adapted from [38].
39
Fig. 8
TE plotted as a function of mole fraction of neutral lipid for the DOTAP/DOPC and the
DOTAP/DOPE systems. DOTAP/DOPC-DNA complexes exhibit the LαC phase. As
described in the text, the same amount of total charged species was used for each TE data
point (which are spread out over nearly four decades) and the membrane charge density of
CL-DNA complexes (σΜ) was varied solely by changing the amount of neutral lipid.
Adapted from [38].
Fig. 9
Typical LSCM images of a mouse L cell transfected with LαC complexes at ΦDOPC = 0.18,
corresponding to cationic membranes with a high charge density σΜ • 0.012 e/Å2 and high
transfection efficiency (cf. Fig. (6)). Red and green fluorescence corresponds to lipid and
DNA labels, respectively; yellow, the overlap of the two, denotes CL-DNA complexes.
Top: The left image is an x-y (top) view at a given z. Arrows in the side views indicate
objects in circles in the x-y plane. The panels on the right are y-z side views along the
vertical dashed lines in the x-y plane. Bottom: The panels on the left are x-z side views
along the horizontal dashed lines in the x-y plane. In the boxes on the right, plots of lipid
and DNA fluorescence intensity along the x-y diagonal or z-axis are shown. Although
the lamellar complexes used here show similarly high TE, no lipid transfer to the cell
plasma membrane is seen in contrast to high-transfecting HII
C complexes (Fig. (5 (A))).
Both released DNA (circle “1”) and intact complexes (circle “2”) are observed inside the
cell. Labels (3) and (4): A complex in the process of releasing its DNA into the
cytoplasm. Corresponding plots of fluorescence intensity as a function of position are
40
shown in boxes in the lower right corner. The cell outline was observed in reflection
mode and therefore appears as blue. Bars = 5 µm for all planes. Adapted from [38].
Fig. 10
Model of cellular uptake of LαC complexes. (a) Cationic complexes adhere to cells due to
electrostatic attraction between positively charged CL-DNA complexes and negatively
charged cell-surface sulfated proteoglycans (shown in expanded views) of mammalian
plasma membranes. (b and c) After attachment, complexes enter through endocytosis. (d)
Only those complexes with a large enough membrane charge density (σM) escape the
endosome through activated fusion with endosomal membranes. (e) Released DNA inside
the cell is observed by confocal microscopy to be present primarily in the form of
aggregates. The DNA aggregates must reside in the cytoplasm because oppositely
charged cellular biomolecules able to condense DNA are not present in the endosome.
Arrows in the expanded view of (c) indicate the electrostatic attraction between the
oppositely charged membranes of the complex and endosome, which tends to enhance
adhesion and fusion. Arrows in the expanded view of (d) indicate the bending of the
membranes required for fusion, which constitutes the main barrier for the process.
Adapted from [38].
Fig. 11
Schematic sketch of an inverted hexagonal CL-DNA complex interacting with either the
plasma membrane or the endosomal membrane �a�. The cell-surface proteoglycans of the
cellular membrane (cf. Fig. (��)) have been omitted for clarity. The outer lipid
41
monolayer covering the HIIC CL-DNA complex has a positive curvature, whereas the
preferred curvature of the cationic lipids is negative as realized in the monolayers coating
DNA within the complex. Thus, the outer layer is energetically costly, which results in a
driving force, independent of the cationic membrane charge density, for rapid fusion of
the HIIC complex with the bilayer of the cell plasma membrane or the endosomal
membrane as sketched in (b). Adapted from [38].
Fig. 12
General structure of ornithine-based multivalent lipids.
Fig. 13
Optical microscopy images of CL-DNA complexes prepared from DOPC, MVL5 and
plasmid DNA in differential-interference-contrast mode (left), DNA fluorescence
(middle, YOYO green) and lipid fluorescence (right, DHPE-Texas Red). Adapted from
[24].
Fig. 14
Small angle XRD patterns of CL-DNA complexes in DMEM for DOTAP/DOPC (top)
and MVL5/DOPC (bottom) lipid mixtures at similar cationic to neutral lipid molar ratios
(31:69 for MVL5; 33:67 for DOTAP). The complexes are in the lamellar LαC-phase. At
the same molar ratio of cationic to neutral lipid, the DNA interaxial spacing is reduced
from 59 Å to 31 Å in the complexes with MVL5, reflecting the higher membrane charge
density. Adapted from [24].
42
Fig. 15
Transfection efficiencies in mouse L cells for cationic lipids DOTAP and MVL5 in
mixtures with DOPC. Adapted from [24].
Fig. 16
General structure of the peptide-lipids. The spacer moiety consists of PEG of variable
length.
Fig. 17
Top: Schematic of a cross-section of a CL-DNA complex containing PEG-Lipids. Gray
circles represent the DNA rods. Bottom: Transfection efficiency of
PEG-lipid/DOTAP/DOPC-DNA complexes (ΦDOTAP = 0.8) as a function of increasing
content of PEG-lipid. The PEG2000-lipid blocks the electrostatic binding of complexes to
cells, whereas the PEG400-lipid has a much weaker effect due to its shorter chain length.
Note the difference in mol-fraction for the PEG-lipids, which correspond to
approximately equal total amounts of PEG.
43
Figures
��������QP
� ��
��
�
�
�
� �
�
�
DNA
���
�
�
5 µm
$
%
& '
Figure 1
44
Figure 2
/αC
45
O
O
PO
O
O
ON
+
O-O
'23&
O
O
O
O
OP
ONH3
+
O O-
'23(
O
O
N+
N
H NH2+
NH2
+
O
NH3+
NH3+
5 Cl�
'263$
Cl�
'27$3O
O
O
O
N+
O
O
N+
OH
'05,(Br
-
Figure 3
46
Figure 4
47
x
y
z
y
x
z
B
x
z
x
y
z
y
Figure 5
48
Figure 6
49
Figure 7
50
F
F
F F
F
Figure 8
51
x
y
z
x
z
y
Figure 9
52
Figure 10
53
Figure 11
54
O
O
O
NO
N +HDG�JURXS
H H
n
Figure 12
55
Figure 13
56
Figure 14
57
50 mol-% 33 mol-% 20 mol-%
0.01
0.1
1
10
100
RLU
per
µg
of p
rote
in /
105
fraction of cationic lipid
09/� '27$3
Figure 15
58
NO
NH
O
O
O
Hn-1
3HSWLGH
Figure 16
59
Figure 17
60
REFERENCES
1 Henry, C.M. &KHP��(QJ��1HZV, ����, ������, 35.
2 Ferber, D. 6FLHQFH������, ���, 1638.
3 Alper, J. 6FLHQFH, ����, ���, 838.
4 Chesnoy, S.; Huang, L. $QQX��5HY��%LRSK\V��%LRPRO��6WUXFW�����������, 27.
5 Friedmann, T. 6FL��$P�� ����, ���, 96.
6 Clark, P.R.; Hersh, E.M. &XUU��2SLQ��0RO��7KHU�, ����, �, 158.
7 Miller, A. D. $QJHZ��&KHP���,QW��(G�� ����, ��, 1768.
8 Felgner, P.L.; Gadek, T.R.; Holm, M.; Roman, R.; Chan, H.W.; Wenz, M.; Northrop,
J.P.; Ringold, G. M.; Danielsen, M. 3URF��1DWO��$FDG��6FL��8�6�$�, ����, ��, 7413.
9 Wolff, J. A. Ed. *HQH�7KHUDSHXWLFV���0HWKRGV�DQG�$SSOLFDWLRQV�RI�'LUHFW�*HQH
7UDQVIHU��Birkhäuser, Boston, ����.
10 Nabel, G.; Nabel, E.; Yang, Z.; Fox, B.; Plautz, G.; Gao, X.; Huang, L.; Shu, S.;
Gordon, D.; Chang, A. 3URF��1DWO��$FDG��6FL��8�6�$�����������, 11307.
11 Extensive and current information on clinical trials in the field of gene therapy can be
found on the internet at http://www.wiley.co.uk/genetherapy/clinical/.
12 Marshall, E. 6FLHQFH, ����, ���, 951.
13 Marshall, E. 6FLHQFH, ����, ���, 510.
14 Rädler, J.O.; Koltover, I.; Salditt, T.; Safinya, C.R. 6FLHQFH�����������, 810.
15 Koltover, I.; Salditt, T.; Rädler, J.O.; Safinya, C.R. 6FLHQFH�����������, 78.
16 Lasic, D.D.; Strey, H.; Stuart, M.C.A.; Podgornik, R.; Frederik, P.M. J. Am. Chem.
Soc., 1997, 119, 832.
61
17 Harrington, J.J.; van Bokkelen, G.; Mays, R.W.; Gustashaw, K.; Williard, H.F. 1DW�
*HQHW�, ����, ��, 345.
18 Willard, H.F. 6FLHQFH�����������, 1308.
19 Koltover, I.; Salditt, T.; Safinya, C.R. %LRSK\V��-�������� ��, 915.
20 Salditt, T.; Koltover, I.; Rädler, J.O.; Safinya, C.R. 3K\V��5HY��/HWW., ����, ����2582.
21 Koltover, I.; Wagner, K.; Safinya, C.R. 3URF��1DWO��$FDG��6FL��8�6�$�����������, 14046.
22 Safinya, C.R. &XUU��2SLQ��6WUXF��%LRO�������� ����440.
23 Schulze, U.; Schmidt, H.-W.; Safinya, C.R. %LRFRQMXJDWH�&KHP�����������, 548.
24 Ewert, K., Ahmad, A.; Evans, H.M.; Schmidt, H.-W.; Safinya, C.R. -��0HG��&KHP�,
����, ��, 5023.
25 Rinehart, J.; Hersh, E.; Issell, B.; Triozzi, P.; Buhles, W.; Neidhart, J. &DQFHU�,QYHVW��
��������������
26 Stopeck, A.T.; Hersh, E.M.; Brailey, J.L.; Clark, P.R.; Norman, J.; Parker, S.E. &DQFHU
*HQH�7KHU����������, 119.
27 Ewert, K.; Safinya, C. R. manuscript in preparation.
28 Farhood, H.; Serbina, N.; Huang, L. %LRFKLP��%LRSK\V��$FWD������������, 289.
29 Hui, S.; Langner, M.; Zhao, Y.; Ross, P.; Hurley, E.; Chan, K. %LRSK\V��-�����������,
590.
30 Lin, A.J.; Slack, N.L.; Ahmad, A.; Koltover, I.; George, C.X.; Samuel, C.E.; Safinya,
C.R. -��'UXJ�7DUJHWLQJ, ����, �, 13.
31 Wrobel, I.; Collins, D. %LRFKLP��%LRSK\V��$FWD������������, 296.
32 Zabner, J.; Fasbender, A.J.; Moninger, T.; Poellinger, K.A.; Welsh, M.A. -��%LRO�
&KHP�, ����, ���, 18997.
62
33 Xu, Y.H.; Szoka, F.C. %LRFKHPLVWU\, ����, ��, 5616.
34 Tristram-Nagle, S.; Petrache, H.I.; Nagle, J.F. %LRSK\V��-�, ����, ��, 917.
35 Felgner, P.L. $GY��'UXJ�'HOLY��5HY�, ����, �, 163.
36 Voet, D.; Voet, J. %LRFKHPLVWU\, Wiley, New York, ����.
37 Bloomfield, V.A. %LRSRO\PHUV������� ��, 1471.
38 Lin, A.J.; Slack, N.L.; Ahmad, A.; George, C.X.; Samuel, C.E.; Safinya, C.R. %LRSK\V�
-�������� ��, 1.
39 Mislick, K.A.; Baldeschwieler, J.D. 3URF��1DWO��$FDG��6FL��8�6�$�����������, 12349.
40 Helfrich, W. =��1DWXUIRUVFK�, ����, ��F, 693.
41 Seddon, J.M. %LRFKLP��%LRSK\V��$FWD, ����, ����, 1.
42 Janiak, M.J.; Small, D.M.; Shipley, G.G. -��%LRO��&KHP., ����, ���, 6068.
43 Gompper, G.; Goos, J. -��3K\V��,, �)UDQFH����������, 621.
44 Wong, G.C.L.; Tang, J.X.; Lin, A.; Li, Y.; Janmey, P.A.; Safinya, C.R. 6FLHQFH, ����,
���, 2035.
45 Remy, J.-S.; Sirlin, C.; Vierling, P.; Behr, J.-P. %LRFRQMXJDWH�&KHP�, ����, �, 647.
46 Kalderon, D.; Roberts, B.L.; Richardson, W.D.; Smith, A.E. &HOO, ����, 39, 499.
47 Dworetzky, S.I.; Lanford, R.E.; Feldherr, C.M. -��&HOO�%LRO������������, 1279.
48 Feldherr C. M.; Akin, D. -��&HOO�%LRO., ����, ���, 1.
49 Smyth Templeton, N.; Lasic, D.D.; Frederik, P.M.; Strey, H.H.; Roberts, D.D.;
Pavlakis, G.N. 1DW��%LRWHFKQRO�����������, 647.
50 Plank, C.; Mechtler, K.; Szoka, F.C.; Wagner, E. +XPDQ�*HQH�7KHU� ����, �, 1437.
51 Bradley, A.J.; Devine, D.V.; Ansell, S.M.; Janzen, J.; Brooks, D.E.�$UFK��%LRFKHP�
%LRSK\V�, ����, ���, 185.
63
52 Lasic, D.D.; Martin, F.J. Eds. 6WHDOWK�OLSRVRPHV, CRC Press, Boca Raton, ����.
53 Allen T. M. 7UHQGV�3KDUPDFRO��6FL�, ����, ��, 215.
54 Martin-Herranz, A.; Ahmad, A.; Evans, H.M.; Ewert, K.; Schulze, U.; Schmidt, H.-W.;
Safinya, C.R.,�submitted to %LRSK\V��-�
Recommended